Top Stories
All Sections
Finance
About
14 results for "vulnerability"
An OpenAI-built AI pipeline produced 948 working Linux kernel privilege escalation exploits in 30 hours for $1.2M — generating an estimated $18.9M–$66.3M in exploit value at historical bug bounty rates.
OpenAI's autonomous AI pipeline analyzed 2,308 Linux kernel bugs and produced 948 working privilege escalation exploits in 30 hours — then launched a patching initiative to fix them.
Google's Gemini 4 Carbon scores 88.2% on SEC-bench Pro for vulnerability identification and 71% on producing working fixes — access restricted to vetted defenders through the Fairwind program.
Anthropic's AI-powered Project Glasswing uncovered 129,000 verified software vulnerabilities in four months, including 33,000 critical or high severity, and is now expanding into a tiered external access program.
Google pre-released Gemini 4 Argon — an autonomous vulnerability-finding and patching model — exclusively to trusted cybersecurity defenders before any public rollout, scoring 68% on the new CWE-bench v1.
Anthropic has publicly released Claude Fable — the renamed Claude Mythos — opening enterprise access to a security AI that autonomously found more than 10,000 high or critical vulnerabilities during its restricted preview.
A new analysis finds 28.3% of CVEs disclosed in 2026 are exploited within 24 hours of publication, driven by AI tooling that converts vulnerability disclosures into working exploits in minutes.
Palo Alto Networks' May patch cycle found 26 CVEs using frontier AI models — versus the typical fewer than 5 per month. The window between vulnerability disclosure and weaponized exploit is shrinking fast.
OpenAI's new Daybreak initiative uses frontier AI models and Codex Security to proactively identify and validate patches for software vulnerabilities — entering the enterprise security market directly.
OpenAI launched Daybreak, a dedicated cybersecurity platform built on GPT-5.5, marking the company's first major product aimed specifically at enterprise security teams.
Google's Threat Intelligence Group has confirmed the first known instance of criminal hackers using AI to discover and weaponize a zero-day vulnerability, marking a concrete shift in the offensive capabilities available to attackers.
Anthropic's Mythos AI alarmed bank executives and forced a White House policy rethink — but cybersecurity researchers say the exploit-generation capability it demonstrated is already widespread across deployed models.
Anthropic launched Project Glasswing, a restricted consortium giving Microsoft, Apple, Amazon, and Google limited access to Claude Mythos — the AI model it is withholding from the public.
AI models have crossed a threshold: they are now competitive with the best human offensive security researchers at finding and exploiting software vulnerabilities.